Cloud-aerosol interactions during autumn over the. Beaufort Sea

Transcription

1 Cloud-aerosol interactions during autumn over the Beaufort Sea James O. Pinto and Judith A. Curry Program in Atmospheric and Oceanic Sciences Department of Aerospace Engineering Sciences University of Colorado, Boulder, CO Janet M. Intrieri NOAA/Environmental Technology Laboratory 325 Broadway, Boulder, CO Journal of Geophysical. Research Special Issue on FIRE-Arctic Clouds Experiment revised 28 March 2000

2 Abstract Cloud and aerosol properties were observed by aircraft in autumn over the Beaufort Sea during the 1994 Beaufort and Arctic Storms Experiment (BASE). The microphysical properties (particle size, concentration, mass, and phase) and vertical structure of autumn clouds are examined as a function of height and minimum in-cloud temperature, T min. Below 2 km, liquid clouds were observed at T min between -5 and -9 C, mixed-phase clouds were observed between -5 and - 20 C and clear-sky ice crystal precipitation was observed at T min as warm as -14 C. Between 2 and 5 km all clouds were mixed-phase and typically consisted of a thin layer of liquid at cloud top with ice extending well below. These mixed-phase clouds were found at T min as low as -32 C. Clouds above 5.5 km were composed entirely of ice at T min as warm as -33 C. The concentration of ice crystals is observed to increase exponentially with decreasing T min. The Arctic atmosphere was relatively clean with condensation nuclei (CN) concentrations rarely exceeding 300 cm -3. The smallest CN concentrations of as low as 50 cm -3 were observed in the boundary layer and just above the surface revealing the importance of precipitation and nucleation scavenging. Thin layers of very large CN concentrations were often observed within and just above low-level clouds. These thin layers may be indicative of aerosol production through gas-to-particle conversion which thrives in the relatively clean and humid conditions often observed at low-levels of the Arctic atmosphere. The vertical variation in aerosol concentrations was consistent with the observed vertical variation in cloud drop concentrations. 1

3 1. Introduction Clouds strongly modulate the surface energy budget over the arctic ice pack. During autumn, as the amount of insolation rapidly decreases, the main influence of clouds is to enhance the emissivity of the atmosphere characterized by extremely small column water vapor amounts. Surface-based observations indicate that clouds are present over the Arctic in autumn 65 to 75% of the time [Warren et al., 1988]. In addition, low-level ice clouds or diamond dust, which are not included in this statistic, become increasingly common as autumn progresses. The microphysical properties of autumn clouds over the Beaufort Sea have not been observed prior to the Beaufort and Arctic Storms Experiment (BASE). A primary objective of BASE was to document and understand the transition in cloud properties over the Arctic ice pack in autumn from predominantly liquid to crystalline [Curry et al., 1997]. The importance of arctic clouds and radiation to the local and global climate has been summarized by Curry et al. [1996]. In view of this importance, recent field campaigns have made intensive observations of arctic clouds and radiation and other aspects of the arctic environment. These include the FIRE (First ISCPP Regional Experiment) Arctic Clouds Experiment [Curry et al., 2000], the Surface Heat Budget of the Arctic Ocean (SHEBA) project [Perovich et al., 1999], and the Atmospheric Radiation Measurement (ARM) Program [Stamnes et al., 1999]. While satellite and surface-based remote sensing provide important information on the climatology and temporal evolution of cloud properties, respectively [e.g., Minnis et al., 2000; Shupe et al., 2000], in situ measurements from aircraft and other platforms are essential for the development and validation of these retrievals. Until the remote sensing measurements are able to retrieve cloud properties to a known accuracy, in situ measurements will remain the most important source of information about cloud microphysical properties. 2

4 The past two decades have seen several aircraft campaigns over the Arctic Ocean that included in situ measurement of cloud microphysical properties. These have included: the June 1980 Arctic Stratus Experiment [Herman and Curry, 1984; Curry, 1986]; the April 1983 and 1986 Arctic Gas and Aerosol Sampling Program (AGASP) [Curry et al., 1990]; the April 1992 and June 1995 observations over the Beaufort Sea by the University of Washington Convair [Hobbs and Rangno, 1998]; the BASE [Curry et al., 1997; Gultepe et al., 2000]; and the April-July 1998 FIRE Arctic Clouds Experiment [Curry et al., 1999; Gultepe et al., 2000; and other papers in this issue]. It has been hypothesized that the microphysical and optical properties of arctic clouds are particularly susceptible to influence by atmospheric aerosol (for an overview, see Curry [1995]). The concentration and size distributions of cloud droplets depend in part on the aerosol in the atmosphere, specifically on the concentration and types of cloud condensation nuclei (CCN) and ice-forming nuclei (IFN) present. The interactions between clouds and aerosol are not simply oneway; scavenging by clouds depletes CCN, but aerosol may also be produced and enhanced within the clouds through chemical and physical processes [e.g., Hegg et al. 1980; Hegg et al. 1990]. Ice particle concentrations in clouds may also be affected by aerosol, although this connection is complex and not well understood. The concentration and composition of IFN are hypothesized to be important for determining the phase of arctic clouds, and therefore their impact on the radiation balance. Aerosol in the Arctic undergoes a profound annual cycle as documented by Barrie and Hoff [1985]. Between January and April, the aerosol concentration is at a maximum owing to long-range transport of mid-latitude pollution aerosol (referred to as "Arctic Haze," for a review see Barrie [1986]). During May and June, the seasonal phytoplankton activity produces a secondary peak in aerosol concentration. Aerosol concentrations are at a minimum during autumn and 3

5 winter due to scavenging by precipitation during the summer month. The aerosol composition varies over the annual cycle as well. The pollution aerosol has a large sulfate component (which serve as CCN) and a significant component of metallic oxides (which serve as IFN). During summer, the aerosol is predominantly sulfate, although soil particles (advected in from snow-free land regions) and biogenic particles are also present. If the aforementioned hypothesis is correct, then there should be a discernible difference in cloud microphysical properties over the course of the annual cycle. The seasonal variation in aerosol properties will be reflected in seasonal changes in droplet concentrations and in temperature-cloud phase relationships. Interannual variability in aerosol sources will play a role in determining cloud properties as well. Curry [1995] compared the microphysical observations of liquid water clouds obtained during the June 1980 Arctic Stratus Experiment [e.g., Herman and Curry, 1984] with earlier observations [e.g., Radke et al., 1976]. They found that the cloud droplet concentrations were significantly larger and droplet radii much smaller than previously observed and they suggested that this may have been caused by increases in anthropogenic pollution. Rathore [1983] observed CCN concentrations in June 1980 that were five times higher than values previously reported by Radke et al. However, Hobbs and Rangno [1998] observed cloud drop concentrations in June 1995 that were much smaller than those observed in June It is now believed that the large droplet concentrations and small droplet radii observed during June 1980 were caused by a sudden increase of natural sulfate aerosols associated with eruptions of Mount St. Helens during May and June, McCormick and Trepte [1987] observed stratospheric aerosol optical depths in the Arctic that were four times the background climatic values for a year after the eruption. Some of this aerosol likely made its way into the troposphere evaluating background natural aerosol concentrations. 4

6 A critical microphysical property of the arctic clouds that determines their radiative properties and hydrological impact is the phase of the condensed cloud water. Clouds are considered to be mixed-phase if ice is observed to be nucleating within a liquid layer. Some of the earliest airborne observations of mixed-phase clouds in the Arctic were made near Barrow, AK in December 1967 by Witte [1968]. He observed mixed-phase clouds at temperatures between -32 and -15 C. Using observation obtained during AGASP flights in April of 1983 and 1986, Curry et al. [1990] reported that most clouds were glaciated at temperatures below -20 C and that mixedphase clouds were present at temperature between -15 and -20 C. Hobbs and Rangno [1998] reported measurements collected in April 1992 and June They observed glaciated low-level cloud layers in April at temperatures as warm as -13 C. Mixed-phase cloud layers were also common at temperature between -20 and -9.5 C. In early June 1995 they observed low-level mixed-phase clouds at temperatures as warm as -4.5 C. Several processes act to determine the phase of clouds. Secondary ice production can elevate ice crystal concentrations well above the observed IFN concentrations. Rangno and Hobbs [1994] determined that the size of droplets in a cloud layer is an important characteristic for the secondary production of ice crystals. Secondary ice production if clouds with large droplets is proposed to have been important in the high ice crystal concentrations observed in low clouds by Hobbs and Rangno [1998]. Mixed-phase clouds may also be found at warm temperatures if the cloud is being seeded with ice crystals originating from above. The concentration of available IFN may also be instrumental in determining the longevity of mixed-phase clouds in the relatively pristine environment in autumn. Pinto [1998] and Jiang et al. [2000] describe the importance of IFN concentrations in maintaining mixed-phase clouds in the Arctic boundary layer. Cloud dynamics also plays a role in determining cloud phase. Gultepe et al. [2000] indicate that vertical 5

7 motion may be important in determining the concentration of ice particles in convective clouds. These studies suggest the importance of several mechanisms in determining the phase of condensed water in Arctic clouds. With the exception of BASE, none of the in situ cloud observations in the Arctic during the past two decades have been made during autumn. A summary of the cloud microphysical and macrophysical properties and aerosol characteristics for the Arctic in autumn is given. This data is put into context through comparison with data collected during past field experiments conducted in other seasons. The impact of cloud-aerosol interactions on arctic cloud microphysics and aerosol profiles is examine and the relationships among various cloud microphysical parameters and to atmospheric conditions are explored. 2. Data and Analysis Method BASE consisted of 14 research flights of the NCAR C-130 aircraft during a 5-week period beginning 21 September and ending 25 October The flights were conducted within the domain shown in Figure 1, with missions flown over sea ice of varying thicknesses and open water. Details of the experiment are given in Curry et al. [1997] and Pinto [1998]. Flight patterns consisted of vertically-stacked, 10-km legs flown at several heights within the cloud layer. Additional l10-km legs were typically flown above and below the cloud layers. Vertical profiles were obtained either through slow ascents/descents either by spiral or more commonly slant profiles. The ascent/descent rate was typically about 5-8 m s -1 so that the profile through a typical cloud layer of 300 m depth covers a horizontal distance of about 4.5 km. The representative scale of an instantaneous (1 sec) measurement during a slant profile is a rectangle 100 m long and 6.5 m high. During BASE, a total of 37 individual cloud layers were sampled at various heights in the 6

8 atmosphere. For determining cloud base and cloud height, the layers were assumed to be independent if they were separated by more than 50 m. The cloud layers are binned into low (i.e., tops below 2 km), mid-level (i.e., top and base both between 1.5 and 5.5 km) and high (i.e., cloud top above 5.5 km). Of the 37 cloud layers observed, 20 were low clouds, 9 mid-level clouds, and 8 high clouds. Each of the high cloud layers were composed entirely of ice while each of the midlevel clouds were mixed-phase. The low clouds were liquid, ice or mixed-phase. 2.1.Instrumentation The NCAR C-130 aircraft was instrumented with a full suite of microphysics probes. The FSSP (Forward Scattering Spectrometer Probe) sizes droplets with diameters ranging from 2 to 47 µm with a bin size of 3 µm. Gardiner and Hallet [1985] discuss the degradation of FSSP measurements in the presence of ice crystals and note that significant overestimates in particle concentration and water content may arise when large aspherical ice particles are present. The 2DC probe sizes particles from 25 to 800 µm in diameter with a resolution of 25 µm, and the 2DP probe sizes particle from 200 to 6400 µm with a resolution of 200 µm. The images obtained by the latter two probes are also used to determine crystal habits. It is noted that the 2DC probe tends to underestimate the concentration of ice particles less than 150 µm in diameter and has an effective detection threshold of 50 µm [Gayet et al., 1993]. Liquid water content (LWC) was measured with the hot-wire King Probe. A Rosemount Icing Rate Detector provided a measurement of the ice rate in supercooled clouds. An algorithm was developed to determine the supercooled LWC from the raw icing rate. The supercooled LWC compare well with the LWC measurements obtained with the KING probe and the FSSP when large ice particles are not present. Aerosol properties were also measured by instruments on the C-130. The aerosol particle 7

9 size distribution was observed with a PCAS probe which detects particles between 0.12 and 3.0 µm in diameter. The size distribution has been corrected to include contributions by coincident particles and corrected sizing of particles at air speeds greater than 50 m s -1 [Baumgardner, 1987]. Because the sample volume for ice particles is indeterminate, this probe performs poorly when high concentrations of ice particles are present [Baumgardner, 1989]. The total aerosol concentration can be estimated with a condensation nuclei counter, which detects Aitken particles larger than 0.01 µm in diameter. A TSI Inc. Model 3760 CN counter was installed after Flight 10 during BASE. 2.2.Cloud Microphysics Analysis Method Determining the microphysical properties of liquid clouds is fairly straightforward. The sensors pertinent to this determination are the King probe, the FSSP, the 2DC probe, and the icing rate detector. All but the 2DC probe give a measure of the liquid water content in supercooled liquid clouds. The FSSP and 2DC probes are used to determine particle size and concentration. Liquid drops between 50 and 100 µm in diameter, as sized by the 2DC probe, are included in the calculations of droplet concentration and effective radius calculation. The effective radius is calculated from the ratio of the second and third moments of the dropsize distribution obtained from the FSSP and 2DC probe. Liquid particles greater than 100 µm in diameter, which were rarely encountered, are not included in these calculations. Problems with one or more of the sensors often arose. Careful comparison of the LWC values obtained with the three sensors allows for determination of the quality of the data. The FSSP data were of poor quality during most of Flights 7 and 14. For these cases, the droplet concentration and size could not be determined. Determination of the microphysical properties in mixed-phase clouds is quite challenging. 8

10 The FSSP probe is sensitive to ice particles; therefore, care must be taken in interpreting FSSP data obtained in mixed-phase clouds. The FSSP probe is known to perform poorly when large ice crystals are present [e.g., Gardiner and Hallet, 1985]. They note that the FSSP particle concentrations can be in error by as much as two to three orders of magnitude in mixed-phase clouds. The LWC measurements made by the King probe and the FSSP within low-level mixed-phase clouds were compared. For typical Arctic clouds observed during BASE (i.e., LWC less than 0.35 g m - 3 ), the FSSP LWC averaged about 10% greater than the King LWC in clouds when ice crystal concentrations were small, increasing to as much as 1000% in clouds with large concentrations of ice crystals. The phase of particles less than 200 µm diameter in mixed-phase clouds is difficult to distinguish. Therefore, within mixed-phase regions of the cloud the concentration of particles between 200 and 800 µm in diameter is used to assess the amount of ice present. The concentration of ice crystals greater than 200 µm in diameter, Ni200, is used to assess the amount of ice present in mixed-phase clouds. The ice concentration, ice water content, and effective radius are calculated using bins from 50 to 800 µm diameter in portions of the profile that are entirely crystalline. Comparisons between ice property calculations within mixed-phase regions using the 200 to 800 µm range and just below mixed- phase regions using the 50 to 800 µm range reveals that the bias in the ice properties is small in the lower portion of mixed-phase clouds. The bias may be larger near cloud top where more small ice crystals are likely present. The IWC is determined from the images obtained with the 2DC probe. The particle volume is determined based on crystal habit, which is determined by the aspect ratio obtained from the 2D images. Ice water contents are determined following the method described by McFarquhar and Heymsfield [1996]. The total ice mass is determined by summing over the 2DC and 2DP size 9

11 distributions, each of which are further binned according to crystal habit. The mass-diameter relationships are a function of habit and the ice density. Assumptions about particle habit and density result in an uncertainty in IWC values of a factor of 2 [Heymsfield, 1990]. Since ice crystals less than 25 µm in diameter are not measured, the mean particle sizes will be overestimated while the total ice particle concentration and ice water content will be underestimated. Attempts to determine ice crystal concentrations with the FSSP probe have had limited success when large ice crystals (i.e., 300 µm diameter) are not present [e.g., McFarquhar and Heymsfield 1997]. However, the FSSP data were not used in this study since most of the ice size distributions had significant concentrations of ice particles greater than 300 µm diameter. 3. Microphysical Properties 3.1.Low-level clouds containing liquid Liquid and mixed-phase clouds were observed below 2 km over the open waters of the Beaufort Sea, the marginal ice zone, and multiyear ice. Several profiles of liquid water content and ice crystal concentration are shown in Figure 2 to illustrate the various vertical structures encountered in low-level clouds. The profiles of LWC fall into one of three categories. Many of the low cloud layers observed consisted of LWC with height to a maximum value near cloud top. The cloud observed during flight 10 is the an exception. Several profiles through low clouds exhibited two distinct LWC maxima within a continuous liquid layer (e.g., Flight 7) or multiple thin liquid layers with several LWC maxima (e.g., Flight 10). The location, time and mean characteristics for 19 different low-level clouds observed during BASE are listed in Table 1. The Table lists the mean properties for each cloud layer. The 10

12 cloud boundaries are determined using a cloud drop concentration threshold of 5 cm -3 after Hobbs and Rangno [1998], or using a King probe LWC threshold of 0.01 g m -3 when droplet concentrations were unavailable. Clouds were found throughout the lowest 1.5 km of the atmosphere with cloud bases typically below 0.6 km. The mean cloud depth for the cases in which both cloud base and cloud top could be determined was 250 m. Clouds were classified as mixed-phase if liquid thresholds described above were reached and the concentration of ice particles greater than 200 µm diameter exceeded 0.01 L -1. Very few cloud layers were observed that were entirely liquid. The minimum in-cloud temperature, T min, for the liquid-only cases ranged between -8.7 and -5.1 C. The cases in which ice crystals originated within the liquid layer are labeled N for ice nucleating cloud layer. The cases labeled P indicate that ice was precipitating into the cloud layer from above. These cases are not used in determining phase-temperature relationships. The low-level, mixed-phase clouds had temperatures ranging between -5.4 C and -20 C. The mean effective radius, r e, for an individual cloud was calculated by weighting the observed r e for a given sub-layer of the cloud with the fraction of the total liquid water path (LWP) in the sublayer. The sub-layer depth is defined by the vertical distance covered by the aircraft in 5 sec. The mean droplet effective radius for cases that had good FSSP data was 9.7 µm. The liquid water path varied between 2 and 70 g m -2 with a mean value of 25.5 g m -2. The LWP is underestimated when cloud base is not reached. Estimates assuming a linear decrease with height of LWC following the observed shape of the LWC profile indicate that the LWP may be underestimated by as much as 25% in some cases. Composite profiles of water content and particle size were calculated from profile data collected in the low-level clouds. Composite LWC and r e profiles for the mostly liquid cloud lay- 11

13 ers (i.e., Ni200 of less than 0.1 L -1 ) are shown in Figure 3 and those for mixed-phase cloud layers (i.e., Ni200 of greater than 0.1 L -1 ) are shown in Figure 4. The maximum LWC observed in these clouds, which occurs at 20% of the cloud depth away from cloud top, is less than 0.25 g m -3. The maximum LWC of the mean profile is about g m -3. The mean effective radius generally increases with height, ranging from 5 µm near cloud base to as large as 14 µm near cloud top. A linear least-squares fit to the data has the mean r e increasing more modestly from 7.5 µm at cloud base to 11 µm at cloud top and a mean value of µm. The composite profiles of liquid and ice water content and particle size for mixed-phase clouds are shown in Figure 4. These clouds have a maximum LWC of less than 0.25 g m -3 with the mean profile having a maximum value of g m -3 at 70% of the cloud depth. The r e profile shows a lot of scatter with values ranging from 5 to 13 µm with droplets begin generally smaller than the mostly liquid clouds. A linear least square fit to the data has r e increasing slightly with height from 7 to 8 µm with a mean value of µm. The IWC increases from 0 near cloud top to 0.03 g m -3 near cloud base with the fraction of the total cloud water mass that is ice increasing to greater than 20% in the lower half of the cloud layer. The ice effective radius shows a great deal of scatter, but there is also evidence of an increase in size toward cloud base. Ice effective radii increases from 150 µm near cloud top to 200 µm below cloud base with the standard deviation being 50µm at each height of the composite profile. Cloud drop concentrations (not shown) were not observed to be a function of height above cloud base in the low-level clouds. There was a great deal of scatter in the droplet concentrations among the different clouds. The median cloud drop concentration for all of the low clouds in which good data is available was 55 cm -3, with most concentrations less than 100 cm -3. Drizzle size drops (i.e., drops with diameter > 200 µm) were seldom observed; however, embryonic driz- 12

14 zle drops (50 to 200 µm in diameter, after Hobbs and Rangno [1998]) were observed in small concentrations (< 4 L -1 ) in a few of the low clouds. In several cases the phase characteristics of the cloud varied significantly over relatively small horizontal distances. An extended leg flown in the cloud layer during Flight 11 reveals mixed-phase and liquid-only regions. The liquid-only regions occur over short segments of the leg (Figure 5). In addition, the concentration of ice crystals varies by several orders of magnitude over very short distances (e.g., between 20 and 25 km Ni200 varies between 0.05 and 7 L -1 ). These horizontal variations obviously have a significant impact on data collected during individual slant profiles through cloud layers. The mean properties of the cloud layer given in Table 1 include data collected during multiple profiles and stacked horizontal legs so that the sample area is typically large enough to adequately characterize the cloud layers. 3.2.Clear-sky ice crystal precipitation Measurements were made in two clear-sky ice crystal precipitation (ICP) layers observed over the multiyear ice pack in mid-october. Analysis of the measurements obtained in one of these ICPs is reported here. The large-scale environment was characterized by a weakening anticyclone and a broadening pool of cold air at 850 hpa. Curry et al. [1997] discuss the thermodynamic processes involved in the formation of this ICP layer from warm moist air originating over the Pacific Ocean. They described the transition from liquid to ice clouds at low-levels as the air mass cools through turbulent and radiative cooling. Profiles of temperature, dewpoint, and frostpoint are given in Figure 6. The ICP layer was contained within a stably-stratified boundary layer (temperature profile is nearly isothermal) over the multiyear ice pack. A shallow, inversion-capped mixed-layer is evident between 1.15 and

15 km. Because the ICP cloud is so optically thin, it could not have contributed enough cloud-top radiative cooling and turbulent mixing to form this layer. It is likely that this mixed layer was formed through radiative cooling atop a shallow mixed-phase cloud layer and subsequent turbulent mixing as demonstrated by Pinto (1998). Temperatures within this ICP ranged between -15 and -12 C, while those in the other ICP case, whose 2D data are not available, are about 4 C colder. Interestingly, these temperatures are warmer than some of the low-level mixed-phase cases described earlier. Both ICP layers were subsaturated with respect to liquid, but supersaturated with respect to ice by more than 15% at times. Vertical profiles of ice microphysical properties are determined from aircraft ascents/descents using 10 s averages, and from vertically-stacked 10-km long horizontal legs. Profiles of ice water content and ice particle effective radius obtained using both methods are given in Figure 7. The depth of the ICP is about 1.5 km, as indicated by the small IWC values measured at this level. Differences between the IWC profiles obtained with the two sampling strategies reveal the importance of defining the spatial scale of the observations. In this particular case, the upper portion of the descent profile was obtained in localized area where IWCs were smaller than those obtained over a broader region as defined by the 10 km leg averages. The particle size generally decreases with height from about 600 µm near the surface to about 100 µm near cloud top. The IWC profile has a concave shape with a maximum of g m -3 near 0.5 km and a rapid decrease with height above 1.0 km. Using the IWC profile obtained from the vertically-stacked legs gives an ice water path of 15.3 g m -2. Some of the microphysical processes occurring in this ICP layer can be inferred from the observed profiles. The minimum size and mass around 1.5 km indicates the height at which the ice nucleation was occurring. Between 0.5 and 1.0 km, the particles are likely growing through 14

16 deposition since both size and mass are increasing in this layer. Below 0.5 km aggregation and accretion of the ice particles are the dominate processes as the particle size increases but the mass is nearly constant. The composite particle size distribution obtained at three heights within the ICP are shown in Figure 8. The FSSP distribution is flat, as is typically observed in ice clouds [Gardiner and Hallet, 1985]. The overlapping portions of the FSSP and 2DC match fairly well; however, the FSSP particle concentrations must be treated with caution since the algorithm designed to relate beam occultation to particle concentration relies on the theoretical Mie-scattering properties of spherical liquid water droplets. Maximum particle sizes as large as 1.5 mm in diameter are observed at the lower levels. The concentration of smaller particles detected by the 2DC probe decreases toward the surface, while the concentration of larger particles measured with the 2DP probe is greatest at 0.46 km. At 1.2 km (Figure 8a) the spectra is fairly flat to about 300 µm, decreasing rapidly thereafter with few particles greater than 800 µm in diameter. The particles at this level exhibit irregular shapes such that no particular habit can be determined (Figure 9a). At 0.46 km (Figure 8b) the ice particle size spectrum is slightly bimodal (this is more evident in the 2DP probe size spectrum), presumably due to aggregation of smaller particles and/or accretion of smaller particles by larger falling crystals. Near the surface (Figure 8c) the concentration of ice particles in each bin has decreased. At low levels, the reduced ice particle concentrations and reduced IWC seen in Figure 7 is due to sublimation which is also indicated by prevalence of ice crystals with rounded edges (Figure 9b). 3.3.Lead-induced clouds In situ observations of a lead-induced condensate plume and its impact on the surrounding 15

17 environment were obtained during BASE on 21 October The lead was located at 74.7 N, 127 W just northwest of Banks Island. Horizontal transects were flown at several levels parallel to the mean wind. The cloud emanating from the open lead is pictured on the cover of BAMS [Vol. 79, No. 2, 1998]. The upwind surface temperature (a proxy for surface air temperature) was -30 C. The open portion of the lead was about 0.5 km wide; however, the wind direction was not perpendicular to the lead so that the fetch exceeded 3 km. The upwind and downwind temperature profiles are given in Figure 10. It is seen that the lead-induced cloud is contained within a shallow layer by a strong surface-based inversion. Despite the shallow depth of the cloud layer, the warming from the lead is seen to penetrate up to 300 m into the inversion layer. A 200 m deep layer of air has been warmed by as much as 1.5 C. This magnitude of warming downwind of an open lead is similar to that observed at a height of 2 m by Ruffieux et al. [1995] and modeled by Glendening and Burk [1992]. In situ microphysical measurements were obtained in the upper portion of the lead-induced condensate plume. Neither the King probe nor the Rosemount Icing Rate Detector detected any liquid water near the top of the plume. Larger particles were detected by the 2DC probe (Figure 11a). They appear to be quasi-circular in shape, possibly frozen droplets with depositional growth, with diameters often exceeding 150 µm. A mean particle diameter of 100 µm is calculated from the particle size distribution (Figure 11b). This value represents an upper limit for the actual mean particle size due to an inability to measure particles < 25 µm in diameter. Most of the ice particles appear to be frozen droplets as the area ratios tend to be greater than 0.7. The total particle concentration is over 50 L -1, which compares well with much earlier measurements made by Ohtake et al. [1982] and is as much as five times greater than the maximum concentration estimated by Andreas et al. [1990]. 16

18 Since the aircraft was unable to fly at altitudes lower than 50 m, we were unable to sample the microphysics and other atmospheric properties at lower levels. However, we found that the cloud was completely glaciated near cloud top at temperatures of -16 to -18 C. The surface air temperature was close to -30 C, as inferred from remote surface skin temperature measurements. The mixing of extreme cold air over the ice pack and relatively warm moist air above the open lead sets up very high supersaturations [Andreas et al., 1981]. The cold temperatures and high supersaturations result in efficient heterogeneous ice nucleation [e.g., Khvorostyanov and Curry, 2000]. Since ice particles were observed less than 1 km downwind of the open lead (giving a transit time of about 4 min), either heterogeneous nucleation of ice or condensation freezing mechanism may have been operating to glaciate the lead-induced cloud. 3.4.Mixed-phase Mid-level Clouds Mid-level clouds are defined here as having both base and top between 1.5 and 5.5 km. Nine different mid-level cloud layers were sampled on four days during BASE. The location, time and mean properties of each mid-level cloud layers are given in Table 2. As in the low-level cloud cases, the cloud boundaries are determined using a cloud drop concentration threshold of 5 cm -3. The concentration of ice crystals does not factor into the determination of the cloud boundaries. All of the mid-level cloud layers encountered during BASE contained at least some ice with most of them having Ni200 greater than 0.1 L -1. Minimum temperatures within the liquid layers ranged between and C. The mean depth of the mid-level mixed-phase cloud layers was 310 m; however, some of the liquid layers were less than 100 m thick. The mean droplet size and LWP of 6.8 µm and 12.3 g m -2, respectively, are significantly smaller than that observed in the low-level clouds. 17

19 Vertical profiles through several mid-level cloud layers are shown in Figure 12. The cloud layers are either individual (e.g., Flight 3C) or part of a multi-layer system of clouds (e.g., Flight 04, 10). Ice in these cloud layers is due to either ice nucleation (labeled N in Table 2) and/or seeding from above (labeled S in Table 2). Examples of both nucleating (Flights 3C and 10) and seeded (e.g., lower cloud layers in Flights 4 and 10) cloud layers are shown in Figure 12. The lower mid-level cloud layers observed during Flights 4 and 10 are clearly being seeded from above. Most of the ice sublimates within 2 km of the mixed-phase cloud base, but sometimes this ice was observed to seed low-level liquid cloud layers. A composite of the vertical profiles of water content, concentration, and particle size for liquid and ice is shown in Figure 13. The range of values is given by the standard deviation in the observations. The LWC and effective radius generally increase with height through the cloud layer. The LWC of these mid-level mixed-phase clouds generally remains less than 0.15 g m -3, with the maximum LWC found at 0.8*cloud depth. The mean LWC profile has a maximum value of g m -2, almost half that observed in the lower cloud layers. The mean droplet effective radius increases from 5.5 µm at cloud base to 8.5 µm at cloud base. The clustering of observed cloud drop sizes around the mean curve indicates that the same dynamic and microphysical processes are operating in most of these clouds. The drop concentration decreases slightly with height through these cloud with the majority of the cloud drop concentrations being between 35 and 80 cm -3. The median droplet concentration of 50 cm-3 is similar to that found in the low-level clouds. The IWC, Ni200, and ice effective radius increase with distance below cloud top to distance below cloud base of one cloud depth. The ice particles tended to be largest at and just below the mixed-phase cloud base and then begin to decrease in size due to sublimation well below cloud base. The non-zero concentration of ice crystals above the mixed-phase cloud top indicates 18

20 that ice is precipitating into the layer from above in some of the cases. The mean concentration of ice crystals in the mixed-phase cloud layer is constant with height at about 0.5 L -1 while the mean IWC increases toward cloud base to a maximum values of about g m -3. Equal amounts of liquid and ice are present at 0.2*cloud depth above cloud base. 3.5.Cirrus clouds Observations were also obtained in several cirrus clouds. Because of the maximum ceiling of the NCAR C-130 aircraft of 7 km, cirrus clouds were under sampled in this study. Cirrus were observed between 4 and 7 km. The mean properties of these clouds are given in Table 3. The clouds were observed at temperature between -44 C and -34 C with the concentration of ice crystals greater than 25 µm in diameter ranging between 30 and 80 L-1. The mean effective radius in these cirrus ranged between 80 and 120 µm. The sizes are biased toward larger particles and the concentrations underestimated since measurements of particles smaller than 25 µm were not made. Liquid was not found in any of the cirrus clouds observed during BASE. 4. Aerosol Properties 4.1.Observations The concentration of aerosols is assessed from measurements obtained with the CN counter and the PCAS probe. The aerosol concentration obtained from the PCAS probe (sizes 0.12 to 3.0 µm) is a subset of the CN concentration which includes all aerosols. The PCAS probe concentration should always be less than the CN concentration. This is generally observed to be true except below precipitating clouds, where the PCAS probe concentrations are larger than the 19

21 CN concentrations. These measurements should be treated with caution and may be explained by the breakup of snow crystals on the inlet of the instrument. The aerosol profiles obtained for nine flights during BASE are shown in Figure 14. The CN concentrations are generally between 100 and 400 cm -3 except in the vicinity of clouds. Aerosol observations obtained with the PCAS probe are generally less than the CN values except below precipitating clouds, where PCAS measurements are erroneous as described above. In many of the profiles the maximum CN concentrations (excluding those related to clouds) occurred aloft (i.e., above 3 km). This illustrates the importance of long-range transport of aerosols into the region occurring aloft, and the importance of scavenging by precipitation in determining the aerosol concentration at the lower levels. The mean background aerosol concentration is determined from the aerosol profiles. The background aerosol concentration is defined as the mean concentration observed between 3 and 6 km outside of liquid cloud layers. The mean background CN and PCAS probe values are given in Table 4. The mean CN concentration varies between 200 and 300 cm -3. The background PCAS particle concentrations are between 30 and 60% of the CN values; however, the PCAS concentrations at low levels, below cloud base, routinely exceed the CN concentrations due to measurement problems. The aerosol profiles indicate that the surface is not a strong source of aerosols. Many of the profiles have their lowest aerosol concentrations just above the surface. The ocean is a potential source for aerosols including sea salt from evaporating spray and the oxidation of dimethylsulfide (DMS) which can lead to the formation of sulfate aerosols [e.g., Ayers et al., 1991]; however, during autumn the exposed open waters begin to freeze up so that the only source of aerosols from the ocean is open leads. The lead that produced a cloud (described in section 3.3) 20

22 was also analyzed for possible contribution to the aerosol budget. CN and PCAS aerosol concentrations were obtained from a leg flown 50 m above and downwind of an open lead. Time series of these data along with the surface temperature which marks the position of the lead are given in Figure 15. The background concentration of CN particles of 50 cm -3 increases by up to a factor of three less than 1 km downwind of the open lead while the PCAS concentrations remain roughly constant. This indicates that the increase aerosol concentration is at sizes smaller than 0.11 µm which may be sulfates forming from DMS. 4.2.Cloud-aerosol Interactions Scavenging by precipitation is thought to be an important sink of aerosols over the Arctic Ocean. At the low temperatures observed during BASE the predominant form of precipitation was snow, an efficient scavenger of aerosols [e.g., Miller, 1990]. Since the PCAS probe data are contaminated in the presence of snow, we rely on the CN measurements to illustrate the importance of scavenging below precipitating clouds. Minimum CN concentrations were often detected below low-level precipitating clouds. This is seen by comparing background CN values with those obtained below cloud base (Table 4) as well as in several individual profiles shown in Figure 14 (e.g., Flights 13, 14, 15, 16 and 18). The minimum CN concentration in these profiles occurs below cloud base, typically just above the surface. Since each of these observations were obtained over sea ice, they indicate that the sea ice surface is probably not an important source of aerosols in the boundary layer. In some cases, the aerosols were scavenged from the atmosphere by cloud processes occurring at or near cloud top. The minima in the aerosol profiles observed at or just above cloud top during Flights 6, 7, and 11 may be related to nucleation scavenging. The broken mixed-phase 21

23 cloud layer observed during Flight 6 (see Table 1) is a good case for studying nucleation scavenging and aerosol processing because profiles were made through both clear and cloudy regions. The profile shown in Figure 14 for Flight 6 was obtained in a cloud free region. The location of the cloud layer observed in surrounding areas is shaded in gray. It is seen that the minimum in the aerosol concentration of this profile corresponds with the location of the cloud layer in surrounding areas. At lower levels, under both cloudy and cloud-free regions, ice crystal precipitation was observed and may have contaminated the PCAS measurements. In many cases, the cloud layer appears to be a source of aerosol particles. Maximum PCAS and CN concentrations in the profiles were often observed within or just above liquid cloud layers. This is seen in profiles obtained during Flights 07, 10, 11, 13, 14, and 18. Oftentimes increased CN concentrations were observed m above cloud top during descent profiles through cloud layers (not shown). The low aerosol concentrations and high relative humidities found in the cloud-free regions of the lower troposphere are ideal conditions for the production of aerosols through gas-to-particle conversion [e.g., Shaw, 1989]. In the profile obtained for Flight 15 the thin layer of high CN concentrations observed just above the ICP layer provides additional evidence that the ICP originated from a liquid or mixed-phase cloud layer that was recently present around 1.8 km. Hobbs and Rangno [1998] observed several ICPs that seemed to be the residuals of shallow liquid cloud layers. This observation and the fact that large CN concentrations were often observed just above cloud top before penetration indicate that the maximum CN concentrations observed in cloud layers is likely not an instrumentation problem but the result of the physical process of gas-to-particle conversion. These aerosol observations indicate that cloud and precipitation processes play an important role in determining aerosol concentration in the lower-troposphere of the Arctic in autumn. 22

24 Very pristine conditions were observed during BASE as the main source of aerosols, long-range transport of pollutants from mid-latitudes, had yet to develop while scavenging by precipitation has been cleansing the atmosphere for several months. It appears that the competing cloud-aerosol processes of nucleation scavenging and gas-to-particle conversion are operating in these clouds. It is unclear from the data if this is a steady state system; however, it is clear the scavenging by precipitation dominates the aerosol budget at the lowest levels of the Arctic atmosphere and that the sea ice surface is not a significant source of aerosols in autumn. 5. Relationships between Cloud Parameters Relationships between various microphysical parameters are explored. In particular, relationships that can be used to parameterize cloud microphysical properties in climate models are developed. Specifically, we examine relationships between mean particle size, concentration, and mass in low and mid-level clouds that contain liquid. In addition, the relationship between ice crystal concentration and temperature is determined and compared with commonly used parameterizations and the dependence of cloud phase on temperature and cloud base height is assessed. The relationships among microphysical parameters are developed using data collected during slantwise ascent/descents through cloud layers. These data include all cloud penetrations regardless of whether or not both cloud top and base were breached. The liquid water content is obtained from the King probe while the effective radius and droplet concentration are obtained from the FSSP data. The relationships among microphysical parameters for low-level clouds are depicted in Figure 16. The cloud liquid water content is seen to increase with increasing drop concentrations. The correlation between these two parameters is significant at The size of the liquid drops is a function of the droplet concentration and the amount of condensed water present. Drop size decreases with increasing droplet concentration and increases with increasing water content (Fig- 23

25 ures 16b & c). These relationships indicate that LWC increases through the activation of additional CCN and the growth of existing droplets. The growth of existing droplets appears to dominate the increase in LWC based on the stronger correlation between re and LWC than between re and Nd. This is in agreement with the limited concentration of aerosols available to serve as CCN at low level described in Section 4. The same analysis is performed for the mid-level mixed-phase clouds. There does not appear to be a relationship between LWC and Nd in these clouds (Figure 17a). Unlike the low-level clouds, mid-level clouds can have very low water contents at large concentrations. On the other hand, the effective radius is a strong function of drop concentration and LWC (Figure 17b & c). The droplet size is a stronger function of concentration than LWC indicating that there is increased availability of CCN at mid-levels of the atmosphere relative to lower levels. Observed Ni200 are related to the minimum in-cloud air temperature in Figure 18. The observations used in Figure 17 were obtained in pure ice and mixed-phase clouds between 100 and 6500 m. Since small ice crystals will grow rapidly to 200 µm when liquid is present it is felt that Ni200 gives an adequate representation of ice crystal concentration in mixed-phase clouds. Ice crystals will grow more slowly in the absence of liquid drops. In the pure ice cases the ice crystal concentration is likely to be significantly underestimated by Ni200. The Ni200 generally increases with decreasing minimum in-cloud air temperature. The curve fit for contact nucleation developed by Meyers et al. [1992] tends to agree with the ice crystal concentrations observed within low-level mixed-phase clouds, while the relationship between temperature and IFN concentration given by Fletcher [1962] agrees better with the ice crystal concentrations observed in mid-level clouds. Both parameterizations tend to overestimate the observed ice crystal concentrations in the cirrus clouds (i.e., T min < 240 K); however, the observed values are expected to be underestimates because the concentration of ice particles smaller than 25 µm in diameter could not be measured. 6. Processes that Determine Cloud Phase 24

26 The production of ice crystals in liquid clouds depends on the droplet sizes, IFN concentration, temperature, seeding by ice crystals from above, and strength of turbulence and entrainment. The importance of each of these processes is explored with the observations obtained during BASE. Hobbs and Rangno [1998] discuss the importance of rime-splintering or the Hallet-Mossop (H-M) process in the production of high ice crystal concentrations they observed in warm clouds over the Arctic Ocean in June. They used a threshold diameter (found by determining the diameter at which the concentration of larger drops exceeds 1 cm -3 ) to determine the importance of the H-M process in producing high ice concentrations in warm clouds. During BASE, the concentration of large droplets was typically too small for the H-M process to be operating. Cloud drop size distributions were determined in several relatively warm low-level clouds. Many of the clouds had very few droplets with diameters greater than 30 µm, as is indicated by the cloud top size distributions shown in Figure 19. In the clouds observed during Flight 11, it does not appear that the droplet size was an important factor in determining the phase. Both clouds observed during Flight 11 contained some ice; however, the cloud with the larger droplets at cloud top contained less ice on average. In addition, the three liquid clouds observed during Flight 10 (Figure 19a) are characterized by temperatures similar to those of Case 11.1 (Table 1) and higher concentrations of droplets larger than 30 µm. The H-M process does not appear to be the most important mechanism determining phase in the clouds observed during BASE. Many of the mid-level liquid cloud layers and a few of the low-level liquid clouds layers have ice precipitating in from above. The impact of this seeding in the mid-level clouds on the cloud phase is difficult to determine; however, we can determine the impact the liquid layer has on the ice originating from above. The ice particles grow rapidly as they fall through a layer of liquid. Despite the thin nature of the mid-level clouds, it is found that the mean ice particle size increases as it passes through the liquid layer. The observed median increase in particle size as it falls through a liquid layer is 28 µm. Assuming a typical fall speed for snow of 1ms -1, the ice particles will have a residence time within the 300-m deep liquid layer of about 5 min. This gives 25